Adaptive Tests of Significance Using Permutations of Residuals with R and SAS

CHAPTER 10

ANALYSIS OF REPEATED MEASURES DATA

10.1 INTRODUCTION

Researchers often collect data on experimental units at several points in time because the additional cost of data collection is small compared to the benefits. For example, individuals who were randomly assigned to a weight loss treatment may be weighed every 2 weeks for 10 weeks. These repeated measurements could be used to evaluate the effectiveness of the treatment over the course of the experiment. By taking measurements every two weeks, instead of one measurement at the end of the 10 week study period, the researchers may obtain much more useful information with little additional cost.

In studies involving animals, researcher often obtain data at regular intervals after an intervention. Because animal experiments are costly and because researchers would like to use few animals in the experiment, they try to obtain as much information as possible from the animals that are used by taking several measurements.

Hence, statisticians often need to analyze data sets that have repeated measurements. This is not always an easy task because the analysis of repeated measurements can become complicated if the difference between the treatment groups varies over time. If the difference does change over time, we conclude that there is an interaction of treatment and time. If the interaction is significant then the analysis proceeds by using the treatment differences at each time, or at a subset of the times. For example, if ...